Text-IF: Leveraging Semantic Text Guidance for Degradation-Aware and Interactive Image Fusion

• DOI: 10.48550/arxiv.2403.16387
• Publication Date: 2024-03-24
• Authors: Xunpeng Yi; Han Xu; Hao Zhang; Linfeng Tang; Jiayi Ma
• Keywords: Degradation

Image fusion aims to combine information from different source images create a comprehensively representative image. Existing methods are typically helpless in dealing with degradations low-quality and non-interactive multiple subjective objective needs. To solve them, we introduce novel approach that leverages semantic text guidance image model for degradation-aware interactive task, termed as Text-IF. It innovatively extends the classical guided along ability harmoniously address degradation interaction issues during fusion. Through encoder decoder, Text-IF is accessible all-in-one infrared visible processing flexible outcomes. In this way, achieves not only multi-modal fusion, but also Extensive experiments prove our proposed strategy has obvious advantages over SOTA performance treatment. The code available at https://github.com/XunpengYi/Text-IF.